We found boys constantly had higher prevalence rates than girls, both in the nine major childhood disabilities respectively and in all disabilities combined. This supports the argument that male is the disadvantaged sex for childhood disabilities. In the U.S.A., in addition to the NHIS,10 the Metropolitan Atlanta Developmental Disabilities Surveillance Program (MADDSP) also studied multiple disabilities in children simultaneously, including ASD, cerebral palsy, ID, VI, and HI.29,30 Among them, ASD, cerebral palsy, ID, and VI were more prevalent in boys, which are compatible with our findings. HI was more prevalent in girls in 1996 (1.2 vs. 1.6 per 1000) but more prevalent in boys in 2000 (1.4 vs. 1.0 per 1000), which compatible with our finding that HI had the smallest sex difference among the five main disabilities studied.
Many factors have been proposed to contribute to the male predominance of childhood disabilities. In term of biological factors, X-linked conditions such as Fragile X syndrome are less often expressed in females than in males.10 In addition, the central nervous system of male infant (and fetus) is more vulnerable to environmental influences such as maternal smoking.6,31 Furthermore, females may have a higher threshold for reaching affectation status than males. For example, female patients of ASD carry a greater heritable mutational “genetic load” than male patients.32
In terms of social factors, in many cultures, boys generally receive greater attention and thus may have a higher rate of being diagnosed when the symptoms and signs are mild.17 In addition, compared to girls, boys are more likely to have behavior problems that may affect others and therefore they are more likely to be referred for services. For example, male patients of ASD show more externalizing behavior problems such as aggressive behavior, hyperactivity, reduced prosocial behavior, and increased repetitive/restricted behaviors and interests, while female patients show more internalizing symptoms such as anxiety and depression.6, 32
We found the disabilities due to primarily neurodevelopmental or mental health conditions had higher male-to-female RRs than those due to primarily physical conditions.28 This was also observed in the patients 6 to 21 years old who were served under Individuals with Disabilities Education Act in the U.S.A. between 2015 and 2016.33 Specifically, the male-to-female ratio was 5.31:1 for ASD, 2.02:1 for SLD, and 1.39:1 for ID (due to primarily neurodevelopmental or mental health conditions) while it was 1.23:1 for VI, 1.14:1 for HI, and 1.12:1 for deaf-blindness (due to primarily physical conditions).33
The etiology of childhood disease could also be largely classified into genetic and environmental causes,34–35 and individual childhood disabilities combined different proportions of genetic and environmental causes. The results of the current study suggested that the larger the genetic contribution (heritability), the higher the male predominance. Among the disabilities we studied, ASD had the highest male-to-female RR, and according to the literature, ASD also had the highest heritability.34–35 Specifically, many family and twin studies estimated the heritability of ASD to be about 38–90%, making it the most heritable of all developmental disorders.34–37 Studies on ID found that genetic causes were present in 25–50% of cases,38–40 and a review of epidemiology studies on childhood HI in the U.S.A. found genetic etiology was reported to contribute 7.2–40%.41 In our study, among these three conditions, ASD had the highest male-to-female RR, and HI had the lowest. Therefore, genetic causes seemed to have larger contributions to the sex differences than environmental causes.
We also found that conditions which are more likely to be diagnosed objectively (with more objective signs and quantitative outcomes) tended to have lower male-to-female RRs. For example, during our study period, HI was diagnosed on the basis of on pure tone hearing test results, which are quantitative outcomes with decibel as the unit,19 and ASD was diagnosed on the basis of the DSM-IV diagnostic criteria8,16 ([1] qualitative impairment in social interaction, [2] qualitative impairments in communication, and [3] restricted repetitive and stereotyped patterns of behavior interests, and activities), which are mainly qualitative outcomes. Specifically, HI (1.08), LFMO (1.12), VI (1.28), and MD (1.34) are more likely to be diagnosed objectively and had the lowest male-to-female RRs, followed by ID (1.37), while SLD (1.57) and ASD (5.88) relied more on subjective symptoms (qualitative outcomes) for diagnosis and had the highest male-to-female RRs. This was also true in terms of male-to-female ratios among the children who were served under Individuals with Disabilities Education Act in the U.S.A.: 1.12:1 for deaf-blindness, 1.14:1 for HI, and 1.23:1 for VI, followed by 1.39:1 for ID, and then 2.02:1 for SLD and 5.31:1 for ASD.33
The fact that boys generally receive more attention, which can lead to a higher diagnostic rate, in traditional Chinese culture is a plausible explanation for the higher prevalence of disabilities in boys.17 If so, with the decreasing number of children given birth and growing awareness of sexual disparity during the study period in Taiwan, the male-to-female RRs should have been decreasing. However, we found the RRs of all disabilities combined increased over time, instead of decreased.
We found male-to-female RRs of all disabilities combined, ID, and SLD decreasing with age between 3 and 17 years in most years, which is consistent with a census based study of ID in U.S.A.42,43 ID is the leading childhood disability and contributed 38.4% of cases of all disabilities combined in 2011,18 and therefore we believe the decreasing trend in male-to-female RRs with age observed in all disabilities combined was mainly due to the contribution of ID. In fact, such a trend was not observed in multiple disabilities (e.g. 16.5% in 2011), which contributed the second largest number of cases, and an increasing trend was observed in ASD, which contributed the third largest number of cases (e.g. 14.4% in 2011). Factors affecting the trend in relation to age may vary across the type of disabilities.
A major limitation of the current study is that the government does not release data on individual cases, and so we could not carry out more detailed analyses. This applies not only to analyses of data on individuals, but also to cross-tabulation of aggregated data. Furthermore, we used prevalence data instead of incidence data, which might affect the assessment of sex differences. Strictly speaking, we were targeting at the etiology of the disabilities, and differences in the incidence are more closely related to the etiology than those in the prevalence, which are also affected by survival.44–46 Nonetheless, because most cases of the disabilities we studied are not fatal but incurable, the relative risks estimated using prevalence data should be very close to those estimated using incidence data. We adopted the “administrative” data from a national registry which obtains the information on qualified persons who applied for the services and thus does not include persons who have not done so.47 Consequently, relative risks estimated using the data might be affected by factors influencing the application. However, because the benefits come with the certification are quite substantial, parents or guardians are unlikely to take different actions simply because of the sex of the children, and so the bias introduced should have been minimized.
Compared to previous studies, the current study has some advantages. We studied nine main childhood disabilities at the same time and made comparisons among them, which makes the inference on etiology more comprehensive. This is evidenced by the findings of the more prominent male vulnerability to disabilities due to primarily neurodevelopmental conditions and disabilities with more genetic contribution. In addition, we used data from the whole country, which are seldom available, and the large number of cases (more than sixty thousand cases in 2011 for example) made it possible to generate stable statistical estimates even after stratification of the population. Nonetheless, all the cases were closely examined and certified by physicians, and therefore the diagnoses are reliable. Furthermore, most previous large-scale or nationwide studies were cross-sectional surveys and collected prevalence data for just one year, but we had data over a period of 12 years and thus were able to assess the time trends. Besides, because the number of cases under three years of age was small, the prevalence data on the age group 3–5 years can be regarded as close estimates of incidence.